I'm considering to use the Coral.AI USB accelerator to add an Edge TPU coprocessor to my Raspberry Pi running OctoPi, enabling high-speed machine learning inferencing using a USB port. It's power performance is sufficient enough:

4 trillion operations (tera-operations) per second (TOPS), using 0.5 watts for each TOPS (2 TOPS per watt)

I'm not happy with Obico and its cloud connections, and I just simply want a model to run locally on the Pi instead. For the purpose of detecting failing or failed prints, spaghetti or other defects. As well as checking if the bed is empty before starting prints. Is there a existing pretrained TensorFlow Lite model to achieve 'spaghetti detector'-like computer vision, locally on a TPU?



You must log in to answer this question.